High performance and low-latency replication using storage mirroring

Abstract

An improved storage replication scheme removes the bottlenecks in the data replication path and allows for high performance replication, both synchronous and asynchronous. The scheme eliminates storage array controllers from the replication data path and provides an implementation of array based replication which can sustain much higher input/output (I/O) write bandwidth with much lower latency from the application's perspective.

What is claimed is: 
     
       1. A system for replicating data in a network, the system comprising a primary storage array with a first array of drives, and a secondary storage array with a second array of drives, the primary storage array and the secondary storage array being coupled to each other via the network, the system comprising:
 a first array controller in the primary storage array receiving and processing an input/output write request from an application server; 
 a first drive controller in the primary storage array having the first array of drives attached to it, wherein the first drive controller receives the processed input/output write request, and writes data in the first array of drives in a distributed manner with each drive of the first array of drives writing a certain portion of the data; and 
 a second drive controller in the secondary storage array having the second array of drives attached to it, the second drive controller receiving a mirroring signal from the first drive controller, the mirroring signal comprising the data and information about the manner of written data distribution in the first array of drives, so that data is replicated to be stored in the second array of drives mirroring the manner of written data distribution in the first array of drives. 
 
     
     
       2. The system of  claim 1 , wherein the second drive controller, though capable of performing RAID and input/output processing independently, operates in a pass-through mode, wherein the second drive controller merely distributes the data in the second array of drives in response to the mirroring signal. 
     
     
       3. The system of  claim 1 , wherein the network has low latency to effectuate drive mirroring at high speed between the primary storage array and the secondary storage array. 
     
     
       4. The system of  claim 3 , wherein the network is one of a storage area network (SAN), a local area network (LAN) or a wide area network (WAN). 
     
     
       5. The system of  claim 1 , wherein the first array of drives contains one or more drives for storing parity data needed for RAID. 
     
     
       6. The system of  claim 5 , wherein the second array of drives contain the same number of drives for storing parity data to preserve the RAID stripe in the data stored in the first array of drives, even though the second drive controller does not independently perform RAID. 
     
     
       7. The system of  claim 1 , wherein the second drive controller transmits metadata information to a second array controller in the secondary storage array. 
     
     
       8. The system of  claim 1 , wherein once the data is mirrored in the second array of drives, the second drive controller sends an acknowledgment signal to the first drive controller. 
     
     
       9. The system of  claim 8 , wherein the data replication process is synchronous. 
     
     
       10. The system of  claim 9 , the first array controller sends a final acknowledgment signal back to the application server upon receiving the acknowledgement signal from the second drive controller to the first drive controller. 
     
     
       11. The system of  claim 8 , wherein the data replication process is asynchronous. 
     
     
       12. The system of  claim 11 , wherein the first array controller sends a final acknowledgment signal back to the application server irrespective of when the acknowledgement signal from the second drive controller is received by the first drive controller. 
     
     
       13. A method for replicating data in a network using a replication system, the system comprising a primary storage array with a first array of drives, and a secondary storage array with a second array of drives, the primary storage array and the secondary storage array being coupled to each other via the network, the method comprising:
 receiving an input/output write request at a first array controller in the primary storage array from an application server; 
 processing the input/output write request and transferring the processed request to a first drive controller in the primary storage array, the first drive controller having the first array of drives attached to it; 
 writing the data in the first array of drives in a distributed manner with each drive of the first array of drives writing a certain portion of the data; 
 receiving a mirroring signal at a second array controller in the secondary storage array from the first array controller, the mirroring signal comprising the data and information about the manner of written data distribution in the first array of drives; and 
 replicating the data in the second array of drives attached to the second array controller, mirroring the manner of written data distribution in the first array of drives. 
 
     
     
       14. The method of  claim 13 , wherein the method further comprises:
 sending an acknowledgment signal by the second drive controller to the first drive controller, once the data is mirrored in the second array of drives. 
 
     
     
       15. The method of  claim 14 , wherein the data replication method is synchronous. 
     
     
       16. The method of  claim 15 , wherein the method further comprises:
 upon receiving the acknowledgement signal from the second drive controller, sending a final acknowledgment signal by the first array controller back to the application server. 
 
     
     
       17. The method of  claim 14 , wherein the data replication method is asynchronous. 
     
     
       18. The method of  claim 17 , wherein the method further comprises:
 sending a final acknowledgment signal by the first array controller back to the application server irrespective of when the acknowledgement signal from the second drive controller is received by the first drive controller. 
 
     
     
       19. The method of  claim 13 , wherein the second drive controller, though capable of performing RAID and input/output processing independently, operates in a pass-through mode, wherein the second drive controller merely distributes data in the second array of drives mirroring the manner of written data distribution in the first array of drives. 
     
     
       20. The method of  claim 13 , wherein the network has low latency to effectuate drive mirroring at high speed between the primary storage array and the secondary storage array, wherein the network is one of a storage area network, a local area network or a wide area network.